Circulating tumor cells are central to metastasis, a particularly malign spread of cancer beyond its original location. While rare, there is growing evidence that the clusters of circulating tumor cells are significantly more harmful than individual cells. Microfluidic platforms constitute the core of circulating tumor cell cluster research, allowing cluster detection, analysis, and treatment. In this work, we propose a new mathematical model of circulating tumor cell clusters and apply it to simulate the dynamics of the aggregates inside a microfluidic channel with the external flow of a fluid. We leverage our previous model of the interactions of circulating tumor cells with varying clustering affinities and introduce explicit bonds between the cells that makeup a cluster. We show that the bonds have a visible impact on the cluster dynamics and that they enable the reproduction of known cluster flow and deformation patterns. Furthermore, we demonstrate that the dynamics of these aggregates are sensitive to bond properties, as well as initialization and flow conditions. We believe that our modeling framework represents a valuable mesoscopic formulation with an impact beyond circulating tumor cell clusters, as cell aggregates are common in both nature and applications.